基于特征增强的非均匀采样SAR三维稀疏成像  

3D sparse imaging for non-uniformly sampled SAR based on feature enhancement

作　　者：孙豆 邢世其[1] 高海峰 庞礴[1] 李永祯[1] 王雪松[1] SUN Dou;XING Shiqi;GAO Haifeng;PANG Bo;LI Yongzhen;WANG Xuesong(State Key Laboratory of Complex Electromagnetic Environment Effects on Electronics and Information System,College of Electronic Science,National University of Defense Technology,Changsha 410073,China;Shaanxi Provincial Geological Survey Institute,Xi’an 710054,China)

机构地区：[1]国防科技大学电子科学学院电子信息系统复杂电磁环境效应国家重点实验室,湖南长沙410073 [2]陕西省地质调查院,陕西西安710054

出　　处：《系统工程与电子技术》2021年第4期901-910,共10页Systems Engineering and Electronics

基　　金：国家自然科学基金(61971429,61901499)资助课题。

摘　　要：对于非均匀采样数据,现有三维稀疏成像方法中的数据局部插值会带来误差,且得到的分布式目标成像结果与其连续散射的本质不符。针对这些问题,首先将成像问题直接建模为三维空间中联合的稀疏重构问题,并通过选取候选散射中心进行字典降维;然后,在降维后的模型中增加三维特征增强约束项,建立三维空间中相邻散射中心之间的联系;最后,结合高斯迭代法以及优化的信号处理技巧,提出了一种高效的模型求解算法。实验结果表明,相比于其他成像方法,本文方法对旁瓣的抑制能力强,成像结果分辨率高、精度高,且保证了分布式目标成像结果的连续性。For non-uniform sampling data,the local data interpolation in the existing three-dimensional(3D)sparse imaging methods will bring errors,and the imaging result of distributed targets is poor.To solve these problems,firstly,the imaging problem is directly modeled as a joint sparse reconstruction problem in 3D space,and the dictionary reduction is carried out by selecting the candidate scattering centers;secondly,the 3D feature enhancement constraints are added to the reduced model to establish the relationship between adjacent scattering centers in 3D space;finally,an efficient model solving algorithm is proposed by combining the Gaussian iterative method and the optimized signal processing scheme.The experimental results show that,compared with other imaging methods,the proposed method has a strong ability to suppress side lobe,whose imaging results are with high resolution and high precision,and performs well on distributed targets.

关 键 词：合成孔径雷达 非均匀采样 三维成像 稀疏重建 特征增强 

分 类 号：TN958.97[电子电信—信号与信息处理]